FACULTY OF SCIENCE AND TECHNOLOGY
MASTER'S THESIS
Study program/specialization:
Master of Science Biological Chemistry
Spring semester, 2019 Open access Writer:
Sunniva Lundal Haaland Faculty supervisor:
Hanne Røland Hagland External supervisor(s):
Eva Bernhoff & Aasmund Fostervold Title of master's thesis:
Characterization of antimicrobial resistance in Norwegian blood culture isolates of Klebsiella pneumoniae using whole genome sequencing.
Credits: 60
Keywords:
Antimicrobial resistance, ESBL, Klebsiella pneumoniae, multidrug resistance
Number of pages: 74 + supplemental material/other: 16
Stavanger, 14.06.2019
Title page for Master's Thesis Faculty of Science and Technology
University of Stavanger – Faculty of Science and Technology Master Thesis
Biological Chemistry, Master of Science Degree Program
Characterization of antimicrobial resistance in Norwegian blood culture isolates of Klebsiella pneumoniae using whole genome
sequencing.
Supervisors:
Aasmund Fostervold Eva Bernhoff
By
Sunniva Lundal Haaland
I
Acknowledgments
I would like to show my gratitude to Stavanger University Hospital, Department of Medical Microbiology for given me this great opportunity.
Thank you to Eva Bernhoff and Aasmund Fostervold for great supervising and advice this past year. Especially thanks to Eva, for all the support and help when it comes to the practical laboratory work. Aasmund has been a great supervisor with the theoretical aspect of this project, and have supported me this past year.
A warm thank you to Ragna, who has been my friend and my go to person when something didn’t quite go as planned. I would like to thank Marit, who has been my friend and helped me throughout the year, especially when it comes to bioinformatics.
Thanks to my friends and family who have supported me this whole year.
Sunniva Lundal Haaland 2019
II
Abstract
Antimicrobial drugs are used to treat and prevent infections caused by microorganisms. For decades, they have saved millions of lives, treated diseases and reduced pain. However, when exposed to antimicrobial drugs, microorganisms develop antimicrobial resistance (AMR) over time. This process is accelerated by the misuse and overuse of antibiotics. To handle this global health threat has become a high priority by WHO.
Klebsiella pneumoniae is a common intestinal bacterium that can cause life-threatening infections, such as pneumonia, wound, soft tissue or urinary tract infections. It is an important reservoir for a number of AMR-genes. K. pneumoniae can acquire extended-spectrum β- lactamase (ESBL) encoding genes which lead to resistance against broad-spectrum
cephalosporins. K. pneumoniae is the ‘K’ in the ESKAPE pathogens, which is the six most significant and dangerous causes of drug-resistant hospital infections identified by the Infectious Diseases Society of America.
In this study, antimicrobial resistance determinants and presence of multidrug resistance in a population of 722 K. pneumoniae isolates were identified using Illuminas MiSeq WGS technology and several bioinformatics tools. It was taken particular interest in third- generation cephalosporins, aminoglycosides, fluoroquinolones, trimethoprim/
sulfamethoxazole and colistin. Detected resistance determinants were compared to detected phenotypical antimicrobial susceptibility, determined by micro broth dilution. This was done for a selection of isolates containing extended-spectrum β-lactamase genes or colistin
resistance determinants.
K. pneumoniae sensu stricto was found to be the most prevalent species in this population of 722 isolates. A high diversity of sequence types (STs) was detected (n=378), where ST107 was the most prevalent (n=67).
20.5% (n=148) of this population had determinants encoding resistance to at least one of the drug classes investigated. Five isolates were found with genotypic colistin resistance due to truncations of PmrB and MgrB. ESBLA-genes were detected in 50 isolates, where blaCTX-M-15
(n=34) was the dominant ESBLA-gene. ST307 containing blaCTX-M-15 was the dominant ESBLA carrying sequence type (n=11).
III 46 isolates were found to be multidrug resistant, mainly to aminoglycosides, fluoroquinolones and trimethoprim/ sulfamethoxazole. However also strongly associated with carriage of ESBLA, predominantly blaCTX-M-15.
Based on the results, genotype cannot reliably predict phenotype for all the tested drug classes. However, presence of ESBLA-genes coincides with phenotypic resistance against third-generation cephalosporins, indicating ability to predict a resistant phenotype.
IV
Abbreviations
AMR - antimicrobial resistance
ESBL - Extended-spectrum β-lactamase K. pneumoniae - Klebsiella pneumoniae
The Norwegian Klebsiella Bacteremiae study – NORKAB MDR – multiple drug-resistant
TMP – Trimethoprim SMX – Sulfamethoxazole PBP – penicillin-binding protein WGS – whole genome sequencing NGS – next generation sequencing ST – sequence type
MBD – micro broth dilution
AST – antimicrobial susceptibility testing MIC – minimum inhibitory concentration MLST – multi locus sequence typing Kbp – kilo base pair
Mbp – mega base pair Bp – base pair
NORM – Norwegian Surveillance System of Antibiotic Resistance in Microbes
V Table of Contents
ACKNOWLEDGMENTS ... I ABSTRACT ... II ABBREVIATIONS ... IV
1. INTRODUCTION ... 1
1.1BACKGROUND ... 1
1.2KLEBSIELLA PNEUNOMIAE ... 2
1.2.1 Klebsiella pneumoniae species ... 2
1.3ANTIBIOTICS ... 3
1.3.1 β-lactam antibiotics ... 3
1.3.2 Aminoglycosides ... 3
1.3.3 Fluoroquinolones ... 4
1.3.4 Trimethoprim/sulfamethoxazole ... 4
1.3.5 Colistin ... 4
1.4ANTIMICROBIAL RESISTANCE ... 5
1.4.1 Development of antimicrobial resistance ... 5
1.5EXTENDED-SPECTRUM Β-LACTAMS (ESBLS) ... 5
1.5.1 Class A ESBL ... 7
1.6EPIDEMIOLOGY OF ANTIBIOTIC RESISTANT KLEBSIELLA PNEUMONIAE ... 7
1.7BACKGROUND METHODS ... 9
1.7.1 Illumina Sequencing ... 9
1.7.2 Antimicrobial susceptibility testing (AST) ... 12
2. AIMS OF STUDY ... 15
3. MATERIALS ... 16
3.1INSTRUMENTS ... 16
3.2COMMERCIAL KITS AND REAGENTS ... 17
3.3COLLECTION OF BACTERIAL ISOLATES ... 18
4. METHODS ... 19
4.1CULTIVATION OF BACTERIAL ISOLATES ... 19
4.2DNA EXTRACTION ... 19
4.3DNA CONCENTRATION MEASUREMENT AND NORMALIZATION ... 20
4.3.1 Concentration measurement ... 20
4.3.2 Normalization ... 21
4.4NEXTERA XT LIBRARY PREPARATION FOR ILLUMINAS MISEQ ... 22
4.4.1 Perform a run using Hamilton ML Star ... 22
4.4.2 Tagmentation ... 23
4.4.3 Library amplification ... 23
4.4.4 Library Clean Up ... 23
4.4.5 Normalize Libraries ... 23
4.4.6 Pool Libraries ... 23
4.5WHOLE GENOME SEQUENCING ON ILLUMINAS MISEQ ... 24
4.6QUALITY MONITORING POST SEQUENCING ... 27
4.6.1 Quality control using MCS during sequencing ... 27
4.6.2 Quality assessment using Sequence Analysis Viewer (SAV) post sequencing ... 27
4.7BIOINFORMATIC ANALYSIS OF SEQUENCE DATA ... 29
4.7.1 Quality check of raw data ... 30
4.7.2 Assembly ... 30
4.7.3 Quality valuation of genome assembly ... 31
VI
4.7.4 Phylogenetic analysis ... 32
4.7.5 Multi locus sequencing typing, species identification and resistance profiles ... 32
4.8ANTIMICROBIAL SUSCEPTIBILITY TESTING ... 33
4.8.1 Micro broth dilution ... 33
5. RESULTS ... 35
5.1RAW DATA AND ASSEMBLY QUALITY ... 35
5.2DISTRIBUTION OF SPECIES, SEQUENCE TYPES AND ESBL-HARBORING ISOLATES ... 36
5.2.1 Identification and distribution of Klebsiella pneumoniae species ... 36
5.2.2 Phylogenetic analysis of the sequenced K. pneumoniae population, multi locus sequence typing and prevalence of ESBL genes ... 36
5.2.3 Distribution of ESBLA genes in dominant ESBLA-gene containing sequence types ... 37
5.2.4 Geographical distribution of isolates with ESBLA encoding genes ... 38
5.3ANTIMICROBIAL RESISTANCE – DETECTION OF GENETIC AMR DETERMINANTS AND PHENOTYPICAL ANTIMICROBIAL SUSCEPTIBILITY TESTING ... 39
5.3.1a Detection of ESBLA encoding genes ... 39
5.3.1b Investigation of imprecise ESBLA encoding genes allele matches ... 39
5.3.2a Distribution of resistance determinants against important antimicrobial drug classes .... 41
5.3.2b Resistance determinant gene variants ... 42
5.3.3 Detection of multidrug resistance ... 43
5.3.4 Comparison of phenotypic antimicrobial susceptibility testing with detection of resistance determinants ... 44
6. DISCUSSION ... 49
6.1DISCUSSION METHODS ... 49
6.1.1 Next-generation Sequencing ... 49
6.1.2 Antimicrobial susceptibility testing ... 49
6.1.3 Bioinformatics analysis ... 50
6.2DISCUSSION RESULTS ... 51
6.2.1 Raw data and assembly quality ... 51
6.2.2 Distribution of species and phylogenetic analysis ... 51
6.2.3 Multi locus sequence typing, ESBLA-harboring sequence types and geographic distribution of ESBLA-genes ... 52
6.2.4 Prevalence of known AMR-determinants ... 53
6.2.5 Genetic characteristics of multidrug resistant isolates ... 54
6.2.6 Comparison of phenotypic antimicrobial susceptibility testing with detection of resistance determinants ... 55
7. CONCLUSION ... 58
8. FUTURE PERSPECTIVES ... 59
9. LIST OF REFERENCES ... 60
APPENDIX A ... 67
APPENDIX B ... 70
APPENDIX C ... 72
APPENDIX D ... 73
1
1. Introduction
1.1 Background
Antimicrobial drugs are used as medicines to prevent and treat infections caused by microorganisms. [1] These medicines have saved many lives by treating infections and reduced pain. [2] When bacteria changes in response to these medicines, resistance to antimicrobial drugs occur. Misuse and overuse of antibiotics accelerate the process for microbes becoming resistant. Antimicrobial resistance (AMR) threatens the effective
prevention and treatment of modern medicine to treat a range of infections caused by bacteria, parasites, viruses and fungi. [1], [3]
AMR is present in every country and is increasingly becoming a serious threat worldwide.
New resistance mechanisms are evolving and spreading globally, which decrease the likelihood for effective treatment of antibiotics. [3] Handling AMR is a high priority for the World Health Organization. In May 2015, at the World Health Assembly the “Global action plan on antimicrobial resistance” was endorsed. The aim is to prevent and treat infectious diseases with effective and safe medicines. [1]
Klebsiella pneumoniae is a common intestinal bacterium that can cause life-threatening infections. [3] K. pneumoniae is the ‘K’ in the ESKAPE pathogens, which is the six most significant and dangerous causes of drug-resistant hospital infections identified by the Infectious Diseases Society of America. [4]
An increase in multidrug resistance (MDR) and production of extended-spectrum β-lactamase (ESBL), has led to the increased resistance against carbapenem antibiotics, which is used as a last-resort treatment of infections caused by MDR isolates. Colistin is the last resort treatment of carbapenem resistant K. pneumoniae. Recently, resistance to colistin have been detected in many countries, making infections caused by this bacterium untreatable. [3]
Whole genome sequencing is a method where entire genomes can be analyzed. The genomic information can be useful in providing knowledge about new resistance strains, dissemination of resistance and regulation mechanisms. [5], [6]
2 1.2 Klebsiella pneunomiae
Klebsiella pneumoniae is a Gram-negative, rod shaped bacterium belonging to the
Enterobacteriaceae family. The bacterium is ubiquitous in the environment and can colonize in animals, plants and humans. [4] K. pneumoniae is an important reservoir for a number of AMR-genes and has been the leading cause of hospital-acquired infections and neonatal sepsis worldwide for many decades. [7] The bacterium is considered an opportunistic pathogen which can be carried asymptomatically in the nose, throat, intestinal tract and skin of healthy individuals. However, it can also cause a range of infections in hospitalized patients, most commonly pneumoniae, wound, soft tissue, or urinary tract infections.
Especially elderly, neonates and immunocompromised are at risk. [5] Strains of K.
pneumoniae is increasingly found to be resistant to multiple classes of antibiotics which makes it difficult to treat infections caused by this bacterium [8]. K. pneumoniae can acquire extended-spectrum β-lactamase (ESBL) encoding genes which lead to resistance against broad-spectrum cephalosporins. It was first reported in 1983 in Germany and has increased ever since. [9] Over the years, there has been an increase of multidrug resistant (MDR) K.
pneumoniae strains, even including resistance to colistin (last line drug). This is identified as an urgent threat to the human health. [10] From 2012 to 2015 in Europe, combined resistance to fluoroquinolones, third- generation cephalosporins and aminoglycosides increased from 17.7% to 18.6%. [11]
1.2.1 Klebsiella pneumoniae species
Three phylogenetically distinct groups within the Klebsiella pneumoniae species, KpI, KpII and KpIII, have now been re-designated as distinct species, K. pneumoniae, K.
quasipneumoniae and K. variicola, respectively. However, recent studies suggest that KpII should be further divided into the two subspecies K. quasipneumoniae subs. quasipneumoniae and K. quasipneumoniae subs. similipneumoniae, KpII-A and KpII-B respectively. All these can cause infections in humans. [4],[5],[12] In this thesis, the term K. pneumoniae will be used when referring to all species, KpI, KpII-A, KpII-B and KpIII, and K. pneumoniae sensu stricto when referring to KpI.
3 1.3 Antibiotics
In 1928, a new era in medicine was launched as Alexander Fleming discovered the first true antibiotic, penicillin [13]. Antibiotics have since then saved millions of lives, treated diseases and reduced pain, which have led to the drastically increase of the human life expectancy [2].
These compounds often act by interfering with different targets in the bacterial cell [14]. β- lactams, aminoglycosides, fluoroquinolones, trimethoprim/sulfamethoxazole and colistin are some of the most relevant groups of antibiotics regarding resistance in Klebsiella pneumoniae and the ones focused on in this study.
1.3.1 β-lactam antibiotics
β-lactam antibiotics are the most widely used antibiotics and includes pencillins, cephalosporins, monobactams, β-lactamase inhibitors and carbapenems. All β-lactam antibiotics have a β-lactam ring in their molecular structure [15], as shown in figure 1. By binding to penicillin-binding proteins and disrupting the synthesis of the bacterial
peptidoglycan cell walls, these drugs are bactericidal. [16]. Production of β-lactamase, followed by alterations in cell wall permeability and extrusion of efflux pumps, are the first mechanism in resistance against these antibiotics in K. pneumoniae. [15]
Figure 1: Molecular structure of penicillins, cephalosporins, monobactams and carbapanems. [17]
1.3.2 Aminoglycosides
Aminoglycosides are powerful, broad-spectrum antibiotics with many useful properties for the treatment of life-threatening infections caused by gram-negative bacteria. [18] They are bactericidal and act by inhibiting bacterial protein synthesis, where binding to receptors on the 30S subunit induce a misread of mRNA. [19] Decreased uptake and/or accumulation of the drug in bacteria, and the bacterial expression of enzymes which modify the antibiotic and thereby inactivate it, are the main mechanisms in resistance against aminoglycosides. [18]
4 1.3.3 Fluoroquinolones
Fluoroquinolones are broad-spectrum antibiotics used to treat infections caused by gram- positive and gram-negative bacteria [20]. They are the only direct inhibitors of the bacterial synthesis and by binding to the enzyme-DNA complex, they stabilize DNA strand breaks created by DNA gyrase (gyrA) and topoisomerase IV (ParC). [21] Additionally, mutations in the quinolone resistance-determining regions of gyrA and ParC genes, plasmid-mediated resistance to quinolones, altered permeability and lower uptake of drug are other mechanisms associated with fluoroquinolones resistance. [15]
1.3.4 Trimethoprim/sulfamethoxazole
The drug combination of Trimethoprim and Sulfamethoxazole (TMP/SMX) is a broad- spectrum antibiotic used for treating infections caused by both gram-negative and gram- positive organisms. It is widely used in treatment of many mild-to-moderate and more serious infections. [22] The drug is bactericidal and works by blocking the two steps in bacterial biosynthesis of essential nucleic acids and proteins. [23]
1.3.5 Colistin
Colistin is a 60-year-old antibiotic with significant activity against Gram-negative bacteria. It was revealed that colistin had severe side effects such as nephrotoxicity and neurotoxicity, and therefore the use of colistin was stopped and replaced by other antibiotics considered safe at the time. The increasing number of infections caused by multidrug resistant bacteria have led back to the use of colistin. [24] It is now being used as a ‘last-line drug’ to treat infections caused by multidrug resistant Gram-negative bacteria. [25] Colistin are bactericidal by disrupting the bacterial outer membrane resulting in cell death. [26] Recently, resistance to colistin has been detected in several countries. [3] In K. pneumoniae, resistance to colistin is related to lipopolysaccharide (LPS) modification following the addition of 4-amino-4-deoxy- L-arabinose to lipid A. [15] Recently, studies focus on the emergence of colistin resistance in multidrug resistant bacteria K. pneumoniae as a result of loss-of-function mutations of the MgrB gene and the discovery of mobile resistance (mcr) determinants. [10], [27]
5 1.4 Antimicrobial resistance
Antimicrobial resistance (AMR) is the ability of microorganisms to resist the effect when exposed to antimicrobial drugs. [3] Bacteria may acquire resistance by de novo mutations or acquire resistance genes from other organisms. [28] The process of bacteria becoming resistant is accelerated by the misuse and overuse of antimicrobials. Microbes that are antimicrobial resistant are indigenous in the environment and can be found in people, plants, animals and food. They can be spread by person-to-person contact, through food and between animals and humans. Proper food-handling, sufficient sanitary conditions, good hygiene and right infection control can slow down the spread of AMR. [3]
AMR threatens the modern medicine in a way it prevents and treats a range of infections caused by bacteria, parasites, fungi and viruses. Without effective antimicrobials, treatment and prevention of infections, surgeries and cancer chemotherapy, for example, would be at high risk due to less effective treatment with antibiotics. [3]
1.4.1 Development of antimicrobial resistance
Infections caused by antibiotic-resistant bacteria are harder to treat than those caused by non- resistant bacteria. [1] In bacteria, resistance can occur in different ways. Horizontal gene transfer allows genes to be inherited or acquired between different species of bacteria.
Mutation can also lead to resistance. [29] Plasmids are small, mobile genetic elements which can add new properties to the bacteria such as resistance against antibiotics. [30]
1.5 Extended-spectrum β-lactams (ESBLs)
β-lactamases are enzymes that degrade the β-lactam ring. This results in a slightly different structure and inactivation of the drug, resulting in resistance to β-lactam antibiotics, such as penicillin and cephalosporins [31]. Extended-spectrum β-lactamases (ESBLs) are plasmid- enclosed enzymes that confer resistance on those β-lactams that were designed to resist such enzyme attack [32]. Plasmids responsible for ESBL production often carry resistance genes against other classes of antibiotics. [33]
β-lactamases can be classified into two major classification schemes, the Ambler and the Bush-Jacoby-Medeiros systems. These systems classify β-lactamases based on their enzyme structure (Ambler) and substrate profile, i.e., which class of β-lactams is degraded and to what
6 degree activity is inhibited by the β-lactamase inhibitor clavulanic acid (Bush-Jacoby-
Medeiros). [34] An enzyme is classified as an ESBL if it is a molecular class A, and a
functional class 2be enzyme. A classification was proposed by Giske et al. which expands the definition of ESBL to other clinically important acquired beta-lactamases with activity against extended-spectrum cephalosporins and/or carbapenems. In this classification scheme, ESBLs are categorized into three classes, ESBLA, ESBLM and ESBLCARBA (table 1). The common ESBL families, CTX-M, TEM, SHV, PER and VEB, belongs in the ESBLA class. AmpC and OXA-ESBL are classified as the miscellaneous ESBLs (ESBLM). Lastly, ESBLs with
hydrolytic activity against carbapenems, the carbapanemases has been designated
ESBLCARBA. This class is further divided into ESBLCARBA-A, ESBLCARBA-B and ESBLCARBA-D
[35].
Table 1: Proposal for classification of class A ESBLs (ESBLA), miscellaneous ESBLs (ESBLM) and ESBLs with hydrolytic activity against carbapenems (ESBLCARBA). [35]
7 1.5.1 Class A ESBL
The most common enzymes in class A ESBL include SHV, TEM and CTX-M.
Since the 1980s, SHV-1 and TEM-2 were the predominantly ESBLs. Early 1990s, a new ESBL family emerged, called the CTX-M group.
In gram-negative bacteria, TEM-1 is the most frequently observed β-lactamase from the TEM family. This enzyme can hydrolyze penicillin and cephalosporins. TEM-3 was the first TEM- type β-lactamase that showed the ESBL phenotype. Most of the new derivatives are ESBLs, however some of these β-lactamases are inhibitor-resistant enzymes. [36]
In K. pneumoniae regarding the SHV family, SHV-1 β-lactamase is most commonly found.
Production of SHV-1 makes K. pneumoniae resistant to ampicillin and carbenicillin. Most of SHV variants have an ESBL phenotype that are characterized by the substitution of a serine for glycine at position 238. Other variants, like those related to SHV-5, have a substitution of lysine for glutamate at position 240. The majority of SHV-type derivatives are ESBLs, but some have an inhibitor-resistant phenotype. [15], [36]
CTX-M hydrolyzes broad-spectrum oximino-β-lactams like cefotaxime, ceftriaxone and aztreonam and are easily inhibited by tazobactam and clavulanate. [37] CTX-M-type enzymes are rapidly spreading among Enterobacteriaceae and more than 200 allotypes are known.
This rapid spread of these enzymes has made them the most prevalent ESBLs in
Enterobacteriaceae, where blaCTX-M-15 being the main enzyme currently encountered in K.
pneumoniae. [15], [38]
1.6 Epidemiology of antibiotic resistant Klebsiella pneumoniae
Antibiotic resistance is present everywhere and can transfer from animals to humans, from food to humans, from person to person and through traveling. [39] Antibiotic resistance in K.
pneumoniae is a public health concern. In the EARS-Net report for 2015, more than one third of the K. pneumoniae isolates were resistant to at least one of the antibiotic groups under surveillance (fluoroquinolones, third-generation cephalosporins, aminoglycosides and
carbapenems). In figure 2, combined resistance to multiple antibiotic groups are shown. [11]
8 Many countries in eastern Europe stand out with 50–<75% of K. pneumoniae strains having combined resistance to third-generation cephalosporins, fluoroquinolones and
aminoglycosides. Resistant isolates in Norway show a percentage of 1–<5%, which is significantly low compared to the rest of Europe. [40]
Figure 2: Antimicrobial resistance of Klebsiella pneumoniae in Europe 2017. Percentage of invasive isolates with combined resistance to third-generation cephalosporins,
fluoroquinolones and aminoglycoside. [40]
However, NORM (Norwegian Surveillance System of Antibiotic Resistance in Microbes) reports an increasing amount of resistant K. pneumoniae since 2000, see figure 3. [41]
Figure 3: Prevalence of non-susceptibility to various antimicrobial agents in Klebsiella pneumoniae blood culture isolates 2000- 2017. TMS*=Trimethoprim/ Sulfamethoxazole [41]
9 1.7 Background Methods
DNA sequencing is used to determine the exact sequence of the nucleotides in a strand of DNA. Since the discovery of Sanger sequencing in 1977, DNA sequencing methods have evolved rapidly over the past decade in what are referred to next-generation sequencing (NGS) techniques. [42]
In both Sanger sequencing and NGS techniques, fluorescent nucleotides are added by DNA polymerase to a DNA strand, where each incorporated nucleotide is identified based on its fluorescent tag. The significant difference between Sanger sequencing and NGS techniques is the number of DNA fragments that can be sequenced at a time. Sanger sequencing are often used for small-scale projects, where the method only sequences a single DNA fragment at a time, while NGS can sequence millions of DNA fragments simultaneously per run. Illumina holds the main part of the NGS marked, although there are some other suppliers. In this thesis, Illuminas MiSeq was used to perform sequencing. [43]
1.7.1 Illumina Sequencing
The workflow on Illumina MiSeq includes four basic steps: 1. Library preparation, 2. Cluster generation, 3. Sequencing and 4. Data analysis.
Nextera XT DNA library preparation protocol, described in this thesis, is the procedure recommended for whole genome sequencing (WGS) of bacterial genomes on Illumina. Other procedures are recommended when sequencing mixed bacterial content or human DNA.
1.7.1.1 Library Preparation
Sample preparation starts with extracted and purified DNA, with a concentration of 0,2 ng/µl.
During the first process called tagmentation, transposomes cut and tag the DNA fragments with adapter sequences. Through library amplification, two sets of unique index sequences are added to each DNA fragment in a process called multiplexing. Libraries with unique indexes can be pooled together and sequenced in the same run. Each read can be identified by the flow cell based on the unique index sequences. P7 and P5 are complementary to the lawn of oligos represented on the surface of the flow cell. In figure 4, an illustration of tagmentation and library amplification is shown. [44], [45]
10 Figure 4: Step A and B – Tagmentation: transposomes cut and attach adapters to the
genomic DNA. Step C – Library amplification: two sets of unique index sequences are attached to the adapters on each end of the DNA fragment forming a DNA fragment ready to be sequenced. [46]
1.7.1.2 Cluster generation by bridge amplification
Clustering is a process where the library is loaded onto the flow cell, a glass plate with lanes coated with two types of oligos. These oligos are complementary to the DNA library
fragments and by hybridization the DNA strands are attached to the flow cell surface. A complement strand of the hybridized fragment is created by a polymerase, and the double stranded molecule is denatured where the original template is washed away. Clusters of 1000 identical DNA copies is formed for each strand through bridge amplification. In this process, the strand flips and forms a bridge by hybridizing to the second type of oligo on the flow cell, shown in figure 5.
11 Figure 5: Bridge amplification creating clonal clusters. Single-stranded DNA molecules bends over and are attached to oligos on the flow cell to form a bridge. [47]
A polymerase extends the hybridized primer forming a double stranded bridge. This bridge is denatured resulting in two copies of covalently bound single stranded DNA templates. Then, the reverse strands are cleaved and washed off, leaving only the forward strands. The free 3’- ends of the forward strands are blocked to prevent unwanted priming. [44]
1.7.1.3 Sequencing by synthesis
Sequencing is a process where a sequencing primer is introduced to the flow cell and is hybridized to the adapter sequence annealing site. Fluorescently tagged nucleotides (dNTPs) are added together with a polymerase. The dNTPs have a terminator that prevents the
polymerase of adding another dNTP. After the addition of the nucleotides the clusters are excited by a light source and a unique fluorescent signal is emitted and detected by the MiSeq instrument. After color detection, the terminator and fluorophore are cut off allowing the addition of a new base call. The length of the read is determined by the number of cycles, which is the incorporation of a nucleotide, base detection and fluorophore/ terminator
cleavage. The base call is determined by the emission wavelength and the signal intensity. In a massively parallel process called Sequencing by synthesis technology, hundreds of millions of clusters are sequenced, shown in figure 6.
12 Figure 6: Sequencing by synthesis technology uses four fluorescent dNTP added together with a polymerase. The clusters are excited by a light source and a unique fluorescent signal is emitted and detected by the MiSeq instrument, which provides a digital image. Data is exported to a text file where a sequence for each read is given. These sequences are further analyzed and aligned during data analysis.
The Illumina MiSeq system provides paired-end sequencing. Meaning that after the first read is completed, the read 1 product is washed away. Then, the index 1 primers are hybridized to the template, and after the index read, the product is washed away. It binds to the second flow cell oligo by flipping over. Then index 2 is read similarly to index 1. A double stranded bridge is created by a polymerase and are denatured leaving only the reverse strands. Read 2 begins, and for both reads, the sequencing steps are repeated until preferred length is
achieved. Then the read 2 product is washed away. [47], [44], [45]
1.7.2 Antimicrobial susceptibility testing (AST)
Antimicrobial susceptibility testing (AST) is performed to determine which antibiotics the bacteria is sensitive to. The information provided from these tests are important to guide antibiotic treatment decisions and predict the outcome. [48] In this thesis, micro broth dilution (MBD) was performed and the results was compared to the detected resistance genes in the dataset provided from sequencing.
13 1.7.2.1 Micro broth dilution
Micro broth dilution is a method used to determine the susceptibility to antibiotics of a bacterial isolate. Mueller-Hinton broth suspension with bacteria is inoculated to plates. Each plate is dosed with antimicrobial agents at appropriate dilutions. All plates include one positive control well where growth is required for the tests to be valid. After the plates have been incubated, quantitative minimum inhibitory concentration (MIC) results can be
determined based on the actual growth of bacteria. The bacterial isolates are categorized qualitatively as either intermediate (I), susceptible (S) or resistant (R) to the different types of antibiotics. [49]
This method was performed using a nephelometer, an automated inoculation delivery system, a digital MIC viewing system along with SWIN software. The MIC-values were interpreted according to MIC breaking points provided by European Committee of Susceptibility Testing (EUCAST). Sensititre NONAG5 and Sensititre NONAG4, shown in figure 7 and 8
respectively, were the plates used in this method. These plates are well suited for testing of Klebsiella pneumoniae and other Gram-negative bacteria, based on the antimicrobial agents dosed in these plates.
Figure 7: NONAG5 plate dosed with antimicrobial agents at appropriate dilutions. [50]
14 Figure 8: NONAG4 plate dosed with antimicrobial agents at appropriate dilutions. [51]
15
2. Aims of study
The main aim of this study is to characterize and define the AMR genes of human blood culture isolates of K. pneumoniae in Norway between 2010 and 2015.
Research questions:
• What is the prevalence of known AMR-genes in a population of 1000 blood culture isolates?
• What are the genetic characteristics of a selected number of multidrug resistant isolates?
• How does presence or absence of genetically detected antimicrobial resistance
determinants correlate with results of phenotypic antibiotic susceptibility testing using relevant methods?
16
3. Materials
3.1 Instruments
Instruments used in this thesis were selected based on the specific methods, see table 2.
Table 2: Instruments used in this thesis to perform following methods. Instruments in green are used to perform sequencing, while instruments in grey are used to perform micro broth dilution.
Instrument Function Supplier City, country
MiSeq [52] High performance
sequencer
Illumina San Diego, CA, USA
Hamilton ML Star [53] Used for automated pipetting of liquid sample material and reagents.
Hamilton Giarmata, Timis County, Romania MagNA Pure 96 [54] High-throughput robotic
workstation for fully automated purification of nucleic acids from up to 96 samples.
LifeScience Roche
Basel, Switzerland
Spark® [55] Multimode microplate reader
Tecan Mannedörf,
Switzerland Sensititre™
Nephelometer [56]
Standardize inoculum density
Thermo Fischer Scientific
Waltham, MA, USA
Sensititre AIMTM Automated Inoculation Delivery System [57]
Doses quickly and accurately 96-well plates
Thermo Fischer Scientific
Waltham, MA, USA
Sensititre™ Vizion™
Digital MIC Viewing System [58]
Automated read of visual results
Thermo Fischer Scientific
Waltham, MA, USA
Sensititre™ SWIN™
Software Epidemiology Module [59]
Generate complete, real- time reports and bar graphs in just minutes
Thermo Fischer Scientific
Waltham, MA, USA
17 3.2 Commercial kits and reagents
Different commercial kits were used for different procedures and machines. In table 3, an overview of the kits used during this study is listed.
Table 3: An overview of commercial kits used in this study. Kits in green are used to perform sequencing, while kits in grey are used to perform micro broth dilution.
Commercial kit Function Supplier City, country
MagNAPure 96 DNA and Viral NA Small Volume Kit [54]
Purifies DNA using magnetic glass particle technology.
LifeScience Roche
Basel, Switzerland
Quant-iTTM dsDNA assay kit, high sensitivity [60]
DNA quantification. ThermoFisher Scientific
Waltham, MA, USA
Nextera XT DNA Library Preparation Kit [61]
Prepare sequencing libraries for small genomes, PCR amplicons, plasmids or cDNA.
Illumina San Diego, CA, USA
PhiX Control V3 [62]
Control library for Illumina sequencing runs.
Illumina San Diego, CA, USA MiSeq Reagent Kit
V3 [63]
Sequencing. Enable the highest output of all MiSeq kits.
Illumina San Diego, CA, USA Sensititre NONAG4
[51]
Plate dosed with antimicrobial agents at appropriate dilutions.
Used when performing
antimicrobial susceptibility test.
Thermo Fischer Scientific
Waltham, MA, USA
Sensititre NONAG5 [50]
Plate dosed with antimicrobial agents at appropriate dilutions.
Used when performing
antimicrobial susceptibility test.
Thermo Fischer Scientific
Waltham, MA, USA
18 3.3 Collection of bacterial isolates
Through the Norwegian Klebsiella pneumoniae bacteremia study (NORKAB), 722 K.
pneumoniae isolates from 17 hospitals in Norway were selected for investigation. K.
pneumoniae NORKAB collected blood culture isolates from adult patients from September 2018 to January 2019. In table 4, an overview of isolates from each hospital is given.
NORKAB investigate the relationship between K. pneumoniae genome, disease severity and outcome, and patient characteristics in patients with K. pneumoniae bacteremia.
Table 4: Number of isolates from different hospitals obtained in this study.
Hospitals Region Isolates
Akershus University Hospital Oslo/ Akershus 97
Sørlandets Hospital South 12
Nordland Hospital HF North 4
Diakonhjemmet West 21
Vestre Viken HF East 46
Hospital Østfold East 11
Finnmark Hospital HF North 4
The University Hospital Nord-Norge North 19
Haukeland University Hospital West 97
Hospital Innlandet East 65
Førde Hospital Trust West 15
Rikshospitalet Oslo/ Akershus 42
St. Olav’s Hospital Middle 39
Stavanger University Hospital West 82
Telemark Hospital South 46
Vestfold Hospital South 69
Oslo University Hospital Oslo/ Akershus 53
Total 17 hospitals 722
19
4. Methods
4.1 Cultivation of bacterial isolates Protocol:
Microbank freezing vials containing Klebsiella pneumoniae were collected from the -80oC freezer. Blood agar plates were labeled with name, number and barcode. A glass bead/magnet was collected from the Microbank vial and transferred to the blood agar plate using an
inoculating loop. The bead was streaked on the plate in a three-dilution pattern as shown in figure 9. The blood plates were incubated overnight at 35oC.
Figure 9: Method for streaking single colonies using an inoculation loop. Numbered lines show the order and direction of streaking with the inoculation loop. The plate to the right shows growth of bacteria after incubation overnight. [64]
4.2 DNA extraction
The bacterial DNA must be extracted before sequencing the genome of the isolates of K.
pneumoniae. To extract DNA, MagNA Pure 96 System is used. The MagNA Pure 96 System purifies DNA, RNA, and viral nucleic acids from a wide range of starting materials using magnetic glass particle technology [54].
20 Protocol for extracting DNA:
A few colonies of Klebsiella pneumoniae were taken out from the overnight blood agar plate and transferred to an Eppendorf tube containing 500 µl of sterile saltwater using an
inoculation loop. The tubes were vortexed and 200 µl of each sample were transferred to separate wells in a MagNA Pure Processing Cartridge. MagNA Pure 96 DNA and Viral NA Small Volume with the Pathogen Universal 200 3.1 protocol was used [65].
4.3 DNA concentration measurement and normalization
All samples must be normalized to the same concentration of 0.2 ng/µl before initiating DNA library preparation.
4.3.1 Concentration measurement
To determine the exact concentration of extracted DNA from the K. pneumoniae isolates, TECAN Spark (2) was used together with Quant-iTTM dsDNA assay high sensitivity kit (1) [66]. Due to high concentrations, the extracted DNA was diluted 1:2, in 10 mM Tris-HCl, pH 7.5, before measuring concentration.
1. Quant-iTTM dsDNA assay high sensitivity kit:
1. The assay components were equilibrated to room temperature.
2. A working solution was prepared by diluting the Quant-iTTM dsDNA HS reagent 1:200 in Quant-iTTM dsDNA HS buffer. For 32 samples (+ 2 standards), 10.4 ml of working solution was made (10.348 ml buffer + 52 µl fluorescent reagent) in a Falcon tube covered with tin foil to prevent daylight from disturbing the fluorescent solution.
3. 200 µl of working solution was loaded to each microplate well.
4. The Quant-iTTM dsDNA HS standards were vortexed and centrifuged before adding 10 µl of each standard to separate wells.
5. 10 µl of each unknown DNA sample was added to separate wells.
6. The microplate was sealed and then placed in the shaker for 1 min at 1400 rpm.
21 2. Measuring DNA concentration using TECAN Spark:
1. The computer and monitor were switched on.
2. The Magellan Spark Control icon was clicked to open the program. The slider opens and the microplate with samples were placed here.
3. From the menu; create a sample ID list was selected. All sample IDs were scanned or typed in. The file was saved on the computer.
4. From the main menu; Start measurement was selected. A method used for Quant- iTTM dsDNA assay high sensitivity kit, called High-Sensitivity was chosen. The correct sample ID list was selected and the method was started.
5. It resulted in a graph and calculated concentration for each DNA sample. This file was saved as a comma-separated values (CSV) file.
4.3.2 Normalization
Hamilton Microlab STAR (ML STAR) was used for automated pipetting of liquid sample material and reagents. [53] The concentration of each sample can be adjusted to 0.2 ng/µl using the program called Normalization on Hamilton.
Perform normalization using Hamilton ML STAR:
1. Hamilton ML STAR and cooling element were turned on.
2. The instrument control computer was turned on.
3. From the ML STAR desktop, Hamilton App Launcher was opened.
4. The program called Normalization was selected.
5. The CSV file from previous method was inserted as a file.
6. The on-screen instructions were followed to load the ML STAR carries. OK was clicked after loading the carriers.
7. OK was clicked to verify all labware positions and the run began.
22 4.4 Nextera XT library preparation for Illuminas MiSeq
The workflow for Nextera XT library preparation is shown in figure 10. Performing Nextera XT on Hamilton is Illumina qualified.
Figure 10: Nextera XT workflow. Safe stopping points are marked between steps.
4.4.1 Perform a run using Hamilton ML Star
1. Hamilton ML STAR, cooling element and instrument control computer were turned on.
2. From the ML STAR desktop, Hamilton App Launcher was opened and the appropriate program, Nextera XT, and the appropriate method, tagmentation, library amplification, clean up libraries, normalization or pooling, was selected.
3. The on-screen instructions were followed to load the ML STAR carries. OK was clicked after loading the carriers.
4. OK was clicked to verify all labware positions and the run began.
23 4.4.2 Tagmentation
This step used the Nextera transposome to tagment gDNA, which is a process that fragments DNA and then tags the DNA with adapter sequences in one step. [67] Tagmentation was performed using Hamilton (as described in 4.4.1), details about the procedure can be found in appendix A.
4.4.3 Library amplification
In this step, the tagmented DNA was amplified using a limited-cycle PCR program. The Index 1 (i7), Index 2 (i5), and full adapter sequences were added to the tagmented DNA by PCR. The adapters and indexes were required for cluster formation. [67] Library
amplification was performed using Hamilton (as described in 4.4.1), details about the procedure can be found in appendix A.
4.4.4 Library Clean Up
In this step, AMPure XP beads was used to purify the library DNA and remove short library fragments. [67] Library clean-up was performed using Hamilton (as described in 4.4.1), details about the procedure can be found in appendix A.
Agilent Technology 2100 Bioanalyzer was used to check the libraries. Agilent High Sensitivity DNA Kit Guide was the protocol used. [68]
4.4.5 Normalize Libraries
The quantity of each library is normalized in this process to ensure more equal library representation in the pooled library. [67] Normalization was performed using Hamilton (as described in 4.4.1), details about the procedure can be found in appendix A.
4.4.6 Pool Libraries
Pooling libraries combine equal volumes of normalized libraries in one tube. [67] Pooling was performed using Hamilton (as described in 4.4.1), details about the procedure can be found in appendix A.
24 4.5 Whole Genome Sequencing on Illuminas MiSeq
Some preparations must be done before the pooled DNA libraries can be sequenced on Illuminas MiSeq. This included diluting and denaturation of the pooled libraries, preparation of the reagent cartridge and the set-up of a run using MiSeq Control Software (MCS). MiSeq Reagent Kit V3, 600 cycles, was used (MS-102-3003, Illumina).
Denaturing and Diluting Libraries and PhiX Control for Sequencing:
1. The incubator was preheated to 98oC
2. 15 µl of the pool was combined with 585 µl of HT1 buffer in an Eppendorf tube. The tube was centrifuged at 280 x g at 20oC for 1 minute.
3. The tube was placed on the incubator for 2 minutes. It was put immediately put on ice for 5 minutes (or until you are ready to load libraries on to the cartridge).
4. 6 µl of 12 pM PhiX Control was added to the libraries.
Preparing the Reagent Cartridge:
1. The cartridge was taken out from the freezer the day before use. One may also thaw the reagent cartridge in water bath for an hour.
2. The reagent cartridge was inverted then times to mix the thawed reagents and it was visually examined that all reagents had been thawed.
3. The foil, labeled Load Samples, on the cartridge was pierced with a 1 ml clean pipette. The total volume of libraries and PhiX (606 µl) were loaded onto the reagent cartridge without touching the foil.
Set Up a Run Using MiSeq Control Software (MCS):
1. From the Home screen, Manage Instrument was selected. A reboot was performed to restart the system.
2. A sample plate was made in the Illumina Experiment Manager (IEM) program.
a. From the IEM main screen; Create Sample Plate was selected.
b. Index kit NEXTERA XT Index kit was selected (24 indexes, 96 samples) c. The plate was given a unique plate name. (e.g. Date_initials)
d. Index scheme was selected – 2 – libraries are dual index.
e. Plate tab was selected. This mimic the layout of a 96-well plate with columns A-H, rows 1-12.
25 f. Sample ID was scanned or typed in along with the correct indexes (each
sample was given a unique pair of indexes).
g. Finish was selected and the plate was saved.
3. A sample sheet was made in the IEM program.
a. From the IEM main screen; Create Sample Sheet was selected.
b. MiSeq was selected.
c. An appropriate application was selected – Fastq only
d. In the Reagent Kit Barcode the reagent kit ID from the reagent cartridge was entered.
e. Index adapter was selected – Nextera XT index kit (24 indexes, 96 samples) f. Index reads, 2 – libraries are dual index was selected.
g. Experiment name, investigator name, description and date were entered.
h. Read type – paired end – was selected.
i. Cycles read: 301 was entered.
4. From the Home screen, Sequence was chosen to set up run.
5. From the Sequence Mode Selection screen, Local Run Manager was selected.
6. From the BaseSpace Options screen; use BaseSpaceTM were selected. Then Next was selected.
7. The sample sheet was selected from the list of runs.
8. Next was selected to proceed to load the flow cell.
Clean the flow cell:
1. A new pair of powder free gloves were put on.
2. The flow cell was removed from the flow cell container.
3. The flow cell was rinsed with laboratory-grade water until both the glass and plastic cartridge were thoroughly rinsed of excess salts.
4. The flow cell was dried with lint-free lens tissue until it was completely dry.
Load the flow cell:
1. The flow cell compartment door was raised, and then the release button to the right of the flow cell clamp was pressed.
2. Holding onto the flow cells edges, it was placed on the flow cell stage.
3. The flow cell clamp was slowly pressed down to close it over the flow cell.
4. The flow cell compartment door was closed. Next was selected.
26 Load PR2 and Check the Waste Bottle:
1. The bottle of PR2 was removed from 2° to 8°C storage. It was inverted to mix and the lid was removed.
2. The reagent compartment door was opened.
3. The sipper handle was raised until it locked into place.
4. The wash bottle was removed and PR2 bottle was loaded.
5. The waste was emptied in an appropriate waste container.
6. The sipper handle was slowly lowered.
Load the Reagent Cartridge:
1. The reagent chiller door was opened. The reagent cartridge was inserted into the reagent chiller until the cartridge stopped.
2. The reagent chiller door was closed followed by the reagent compartment door. Next was selected. Start run was selected when the machine was ready.
Perform a Post-Run Wash:
1. When the run was completed, the Next button appears. Next was selected to perform a post-run wash.
2. A fresh solution with Tween 20 and laboratory-water was made:
a. 5 ml 100% Tween 20 was added to 45 ml laboratory-grade water. These volumes resulted in 10% Tween 20.
b. 25 ml 10% Tween 20 was added to 475 ml laboratory-grade water. These volumes resulted in a 0.5% Tween 20 wash solution.
c. The solution was inverted five times to mix.
3. The wash component was prepared with a fresh wash solution:
a. 6 ml of fresh wash solution was added to each reservoir of the wash tray.
b. 350 ml was added to the wash bottle.
4. When the run was completed, start wash was selected.
5. The reagent compartment door and the reagent chiller door was opened and the reagent cartridge was replaced with the wash tray.
6. The PR2 bottle was replaced with wash bottle. The waste bottle was removed and its content discarded appropriately. It was then placed back to the reagent compartment.
7. The sipper handle was lowered and the compartment door was closed. Next was selected and the wash started.
27 4.6 Quality monitoring post sequencing
4.6.1 Quality control using MCS during sequencing
When sequencing using MCS, quality statistics to monitor parameters were provided during the run. Parameters such as Cluster density, Cluster passing filter %, quality score (%Q30) and estimated yield in megabases (Mb) were shown. During and after each run, these parameters were inspected. In table 5, specifications expected when using MiSeq V3 Reagent kit is given [69].
Table 5: Specifications when using MiSeq V3 Reagent kit. Expected value and a description is given for each quality statistics.
Quality statistics Expected value Description
Cluster density 1200-1400 K/ mm2 Number of clusters per square millimeter on flow cell (K/ mm2) Cluster passing filter
% (CPF%)
As high as possible The percentage of cluster passing the Illumina chastity filter.
Quality score, %Q30 > 70% bases higher than Q30 at 2 × 300 bp
Average percentage of bases
>Q30. Q30 = one base call in 1000 is predicted to be incorrect.
Output 13.2-15 Gb The amount of output per flow
cell.
4.6.2 Quality assessment using Sequence Analysis Viewer (SAV) post sequencing
When a run is finished, the output directory was opened in Sequence Analysis Viewer (SAV).
Under the folder Charts, 1) Flow cell chart, 2) Data By Cycle, 3) QScore Distribution, 4) Data By Lane and 5) QScore Heatmap were shown. See figure 11 for an overview of the Chart folder.
28 Figure 11: Screenshot of a run from Sequence Analysis Viewer (SAV). Diagrams/ scales can be analyzed and viewed.
1) The Flow Cell Chart shows color-coded graphical quality metrics per tile for the entire flow cell. This was used to judge local differences per cycle, per lane or per read in
sequencing metrics on a flow cell.
2) The Data By Cycle plot shows the progression of quality metrics during a run as a line graph. This was used to judge the progression of quality metrics during a run on a cycle by cycle basis. Quality metrics used in this study: [70]
• %Q30: The percentage of bases with a quality score of 30 or higher. See figure 12.
• %Base: The percentage of clusters for which the selected base (A, C, T, or G) has been called.
• Error rate: The calculated error rate, as determined by a spiked in the PhiX control sample.
29 Figure 12: Screenshot from a run in Sequence Analysis Viewer. This shows the Data By Cycle plot with chart %Q>=30.
3) The QScore Distribution plot shows a bar graph that illustrates the number of bases by quality score. This was used to judge the QScore distribution for a run, which is an excellent indicator for run performance.
4) The Data By Lane plots shows quality metrics per lane. This was used to judge the difference in quality metrics between lanes.
5) QScore Heatmap is a heatmap of Q-scores. It shows a quick overview of the Q-scores over the cycles. This was not looked much into, except if the scores were very different than expected.
[70]
4.7 Bioinformatic analysis of sequence data
Raw data from the MiSeq run were stored on the MiSeq instrument under the folder output.
This data consists of short-read sequences, sequence identifiers and quality scores stored in FASTQ format.
30 4.7.1 Quality check of raw data
FastQC v.0.11.7 [71] were used to evaluate the quality of the short-read sequences. FastQC provides a modular set of analyses to check if the data has any problems before doing further analysis. The program provides an output report for each sequenced isolate, showing negative or positive results within parameters such as base statistics, per base sequence quality, per sequences quality score, per base sequence content, per base GC content, per sequence GC content, per base N content, sequence length distribution, sequence duplication levels, overrepresented sequences and Kmer content. [72]
Another program that was used in addition to FastQC is MultiQC v1.4 [73]. MultiQC is a tool that aggregates all the gathered FastQC reports into one single report. This provides more overview of the samples, which makes it easier to analyze and sort out samples that are not within the optimal parameter ranges. [74]
Quality and adapter-based read trimming:
Trim Galore v0.6.1 [75], which uses CutAdapt and FastQC to apply quality and adapter trimming to FASTQ files, was used to trim the raw reads. It trims 1bp off each read at the 3’- end, trims low-quality bases (<Phred score 20) at the 3’-end, it removes any adapter
sequences at the 3’-end and lastly removes read-pairs that are less than 20 bp long. Phred 20 indicates that 1/100 can be wrong.
4.7.2 Assembly
To reconstruct the genome from the raw data de novo assembly was performed.
Sequence reads were assembled into contigs, a stretch of continuous sequence generated by overlapping sequence reads [45]. See figure 13 for an illustration of the process. The coverage quality of de novo sequence data depends on the size and continuity of the contigs, and the sequencing depth [76].
De novo assembly was performed with Unicycler v0.4.4 [77], an assembly pipeline for bacterial genomes [78]. Unicycler uses SPAdes v3.13.0 [79], an assembly toolkit containing various assembly pipelines, and Pilon v1.22 [80] for assembly polishing.
31 Figure 13: De novo assembly. Overlapping short reads are aligned to make continuous sequences called contigs.
4.7.3 Quality valuation of genome assembly
Quast v.4.6.3 [81] was used to evaluate the quality of the genome assembly. Several metrics are provided by Quast [82] (ideal quality):
• Number of contigs and length of contigs (<700 and long)
• Length of the largest contig (>0.135 Mbp)
• N50 – the length of the collection of all contigs of that length (or longer) that covers 50% of the assembly (minimum 30 000bp)
• L50 – number of contigs that are equal to or longer than N50 (as few as possible)
• GC% (57% expected for K. pneumoniae)
Quality check of sequencing depth:
The trimmed FastQ files were aligned against their respective assembled contigs with Seqdepth v1 (https://github.com/marithetland/SeqDepth), which uses the Burrows-Wheler Alignment v0.7.17-r1188 (BWA-MEM) [83] and Pickard Tools v2.17.8 [84] to create a Binary Alignment Map (BAM). It then uses SAMtools v1.7 [85] to calculate the overall sequence depth. SAMtools determines the depth of each position and average sequence depth is determined by dividing the average of all positions by the total genome size.
32 4.7.4 Phylogenetic analysis
RedDog v1beta.10.3 [86] is a comparative analysis pipeline that uses high-throughput
sequences for large numbers of bacterial isolates [87]. To evaluate the relationship of the 722 isolates, the RedDog pipeline was used with the raw reads as input and the well characterized Klebsiella pneumoniae strain HS11286 (Genbank accession: NC_016845.1) as the reference genome.
Bowtie2 v2.2.5 [88], with the setting ‘sensitive local-mapping’, was used by RedDog to map the isolates’ reads against the reference genome. SAMtools v1.7 [89] was used to identify single-nucleotide variants (SNVs) and FastTree created a maximum likelihood tree of the aligned isolates. To visualize the core genome SNV tree, msctrees-main.R
(www.github.com/marithetland/msctrees/) was used.
4.7.5 Multi locus sequencing typing, species identification and resistance profiles
Kleborate v0.3.0 [90] is a software that identifies emerging pathogenic Klebsiella pneumoniae lineages, monitors antibiotic resistance and looks out for the convergence of antibiotic
resistance and virulence. It screens Klebisella genome assemblies for multi-locus sequence types (STs), species and antibiotic resistance genes, including mutations leading to colistin or fluoroquinolone resistance. Kleborate uses Mash [91] to compare the assembly to a curated set of Klebsiella assemblies from NCBI (https://www.ncbi.nlm.nih.gov/assembly) and reports the species with the closest match. Kleborate was used to screen the sequenced Klebisella pneumoniae collection with the assembled contigs as input.
BLAST+ [92] was used to assess ESBL-encoding genes where Kleborate reported imprecise allele matches or incomplete coverage.
Kleborate determines MLST based on the K. pneumoniae MLST scheme hosted at institute Pasteur MLST system (https://bigsdb.pasteur.fr/klebsiella). Three isolates did not match any ST in that database and where therefore submitted to the institute Pasteur MLST curators and were assigned new ST types.
33 4.8 Antimicrobial susceptibility testing
From the population of 722 K. pneumoniae, a collection of 41 isolates were chosen for phenotypical characterization by antimicrobial susceptibility testing (AST). Broth microdilution was the method performed.
4.8.1 Micro broth dilution
Micro broth dilution (MBD) is a common method to determine the susceptibility of bacteria to antibiotics, both qualitatively and quantitatively. Sensititre™ Nephelometer, Sensititre AIMTM Automated Inoculation Delivery System, Sensititre™ Vizion™ Digital MIC Viewing System and Sensititre™ SWIN™ Software Epidemiology Module, all provided by Thermo Scientific, were used in this method. [57] A collection of 41 isolates were chosen from the population of 722 Klebsiella pneumoniae isolates. The selected isolates tested in this thesis consist mostly of ESBLs isolates and isolates with possible resistance against colistin.
Protocol:
1. Preparation of 0,5 McFarland inoculum suspension
a. 3-5 colonies from an overnight culture on a non-selective blood agar were taken out using a sterile loop.
b. The colonies were suspended in SensititreTM destilled water and mixed. The density was measured to 0,5 McFarland using a nephelometer.
2. Preparation of Mueller-Hinton broth dilution
a. 10µl of 0,5 McFarland suspension was added to 11ml of SensititreTM Cation adjusted Mueller-Hinton broth w/ TES and mixed well.
3. Dispension of Mueller-Hinton suspension into the Sensititre NONAG4 and Sensititre NONAG5 plates using Sensititre automated inoculation delivery system (AIM)
a. A Sensititre AIM was used to dispense 50µl inoculated broth into all the 96 wells on each of the micro titer plate containing different antimicrobial agents in different concentration.
b. 1µl from the well with positive control was spread on to a blood agar plate.
This colony count was done to see if the Mueller-Hinton suspension had a satisfactory concentration.
4. Incubation
a. The inoculated plate was sealed and incubated for 18 hours at 35oC.
34 5. MIC-value determination using the Vizion plate reader.
a. After the incubation, the positive control agar plates were inspected. 50-100 colonies were expected.
b. The plates were read using the Vizion plate reader and Swin software. The MIC-values for each antimicrobial agent were set by marking the well with the lowest concentration that showed no growth. The software provided the MIC- values and resistance characteristics (susceptible, intermediate and resistant) according to EUCAST clinical breaking points.
35
5. Results
5.1 Raw data and assembly quality
The quality of the raw data output was analyzed with FastQC and the assembly quality with Quast. FastQC provided good quality scores for the majority of isolates, however, 20 isolates were re-sequenced due to poor quality score (see appendix B for details). After resequencing, all isolates (n=722) showed satisfactory qualities (table 6), see method section 4.7.3 for ideal qualities. GC% score ranged between 57-58.4%. Average sequence depth was 58.47.
Table 6: Quality metrics for the 722 sequenced isolates.
Quality metric Average Standard deviation Range
Number of contigs [n] 114 51.5 36-331
Largest contig [Mbp] 0.58 0.24 0.13-1.9
Total length [Mbp] 5.4 0.17 4.9-6
GC score [%] 57.45 0.19 57-58.4
Sequencing depth [X] 58.47 18.39 16.39-115.6
N50 ranged from 34.9-982.1 kbp and L50 ranged from 2-45. The higher score of N50, the lower score of L50 (figure 14), i.e. the longer the contig, the fewer contigs. The majority of isolates laid within a range from L50=5-20 and N50=70-400 kbp.
Figure 14: The correlation between L50 and N50.
0 5 10 15 20 25 30 35 40 45 50
0 200000 400000 600000 800000 1000000 1200000
L50
N50
36 5.2 Distribution of species, sequence types and ESBL-harboring isolates
5.2.1 Identification and distribution of Klebsiella pneumoniae species
Among the 722 Klebsiella pneumoniae isolates, four different species of Klebsiella were detected (table 7). K. pneumoniae sensu stricto was the most prevalent species (n=566), followed by K. variicola (n=120). 13 K. variicola, 2 K. quasipneumoniae subs.
quasipneumoniae and 1 K. pneumoniae sensu stricto isolates showed a weak species match (see method section 4.7.5).
Table 7: Distribution of species from the population of Klebsiella pneumoniae.
Species Proportion of
strains
Number of strong match isolates
Number of weak match isolates Klebsiella pneumoniae sensu stricto 78% (n=566) 565 1
Klebsiella variicola 17% (n=120) 107 13
Klebsiella quasipneumoniae subsp.
similipneumoniae
3% (n=24) 24 -
Klebsiella quasipneumoniae subsp.
quasipneumoniae
2% (n=12) 10 2
Total 100% (n=722) 706 16
5.2.2 Phylogenetic analysis of the sequenced K. pneumoniae population, multi locus sequence typing and prevalence of ESBL genes
A core chromosomal single-nucleotide variant (SNV) tree was made to visualize the phylogenetic relatedness of K. pneumoniae species. This tree (figure 15) divided K.
pneumoniae sensu stricto, K. variicola, K. quasipneumoniae subsp. similipneumoniae and K.
quasipneumoniae subsp. quasipneumoniae into different branches.
A high diversity of sequence types (STs) was discovered, with 378 different STs among the 722 isolates. STs occurring 1 or 2 times made up 60% of the isolates. In figure 15, STs that occur 5 or more times are highlighted using different colors. ST107 was detected as the most prevalent ST (n=67). Three new STs were detected and thus assigned 4009, 4010 and 4011 by the Institute Pasteur, who provides genotypic data for K. pneumoniae isolates based on
MLST. The ESBL genes were more frequently detected in ST307 and ST45 than other STs.
37 A total of 50 ESBLA-harboring isolates were detected in this population. One isolate
containing an ESBLA-gene was found among K. quasipneumoniae subsp. similipneumoniae, whereas all the other ESBLA-genes were detected in K. pneumoniae sensu stricto.
Figure 15: A core chromosomal SNV tree highlighting species, ESBL and the most prevalent sequence types.
5.2.3 Distribution of ESBLA genes in dominant ESBLA-gene containing sequence types STs that occurred 5 or more times with ESBLA-genes are shown in table 8. ST307, which harbored blaCTX-M-15, was found to be the most prevalent sequence type (n=11/50, 22%) among the ESBLA-harboring isolates. All ST307, except one, were found with blaCTX-M-15. Less than 1/3 of ST45 were found with an ESBLA-gene.
